Settings
Configure API keys, LLM providers, and app preferences.
Opening Settings
Click the gear icon in the header toolbar to open the Settings panel.
AI Configuration
To use AI-powered features like natural language pipeline input and intent-based expressions, you need to configure an API key.
Setting Up Your API Key
- Open Settings via the gear icon.
- Select your AI Provider from the dropdown (OpenAI, Anthropic, Google, and others are supported).
- Choose a model from the available options for that provider.
- Enter your API key in the key field.
- Click Test next to the key field to verify that it works.
- Click Save to store the settings.
Supported Providers
| Provider | Example Models |
|---|---|
| Anthropic | Claude Opus 5, Claude Sonnet 5, Claude Haiku 4.5 |
| OpenAI | GPT-5, GPT-5 Mini, GPT-4o |
| Gemini 2.5 Flash, Gemini 2.5 Pro | |
| xAI | Grok 4, Grok 3 Mini |
| Mistral | Mistral Large, Mistral Small, Codestral |
| DeepSeek | DeepSeek Chat, DeepSeek Reasoner |
| Groq | Llama 3.3 70B, Llama 3.1 8B |
| OpenRouter | One key routes to hundreds of models |
| Kimi (Moonshot) | Moonshot v1 |
| Ollama | Any local model, no key needed, nothing leaves your machine |
The model dropdown starts with a few suggested models per provider, but you rarely need to know ids by heart: click Load available models next to the Model label and the dropdown fills with every model your API key can actually use, fetched live from the provider. For local servers the list loads automatically and shows exactly what you have installed. A free-text Type a model id… option remains for anything else, with autocomplete from the loaded list.
Local models (Ollama, LM Studio, llama.cpp, vLLM)
Pick Ollama / local server as the provider and the app talks to a model server on your own machine. No API key, and nothing leaves localhost. The Server URL field defaults to Ollama's http://localhost:11434; point it at any OpenAI-compatible server instead, such as LM Studio's http://localhost:1234.
Ollama only accepts browser requests from origins it trusts, so allow this site once when starting it:
OLLAMA_ORIGINS="https://exploremydata.com" ollama serveTwo limits worth knowing: the connection must be to localhost (machines elsewhere on your network are blocked by the site's security policy), and Safari blocks pages on https from calling localhost at all, so use Chrome, Edge, or Firefox for local models.
Privacy & Security
API keys are stored per provider in your browser's IndexedDB only, and switching providers keeps each key in place. Keys are never sent to ExploreMyData servers, and each key is only ever sent to its own provider. When you use an AI feature, the request goes directly from your browser to the LLM provider's API endpoint. No data passes through any intermediary.
Appearance
Dark / Light Mode
Toggle between dark and light mode using the theme button in the header. The app defaults to your system preference on first load, but you can override it. Your choice is saved and persists across sessions.
Clear All Data
The Settings panel includes a Clear All Data button, under Browser Storage, that resets the app to its initial state after you confirm. This removes:
- All uploaded files from the in-browser DuckDB database.
- All pipelines and view states.
- Your stored API key and provider settings.
Use this if you want a fresh start or need to free up browser storage.
Tips
- You only need an API key if you want to use AI features. All other operations work without one.
- If the Test fails, double-check your API key and make sure it has not expired.
- See Using AI Features for a full guide on what you can do with AI.